
Skill Contract Reviewer
Checks an Agent Skill folder's SKILL.md structure and suggests the right skill primitives before building
- Repository
- rohitg00/ai-engineering-from-scratch
- Stars
- 76,331
- Licence
- MIT
- Files
- 5 files
What it does
When installed, this skill makes an AI agent check a proposed Agent Skill folder before it is turned into a reusable skill. The agent reads two reference notes, runs a small read-only Python script that validates the SKILL.md header fields and structure, then reads the JSON report and recommends the smallest set of skill building blocks (instructions, capabilities, or lifecycle steps) for the job.
Great for
- Developers packaging a repeated workflow into a SKILL.md skill for Claude or another agent
- Teams reviewing whether a proposed skill is well-formed before sharing it
- Authors deciding whether a task needs instructions, a tool, or a scripted step
Skip it if
- Anyone expecting a full security audit: the script only checks the core identity fields and file structure, not what the skill's other instructions or scripts do
- Non-technical users who have no reason to create skills, since the output is aimed at skill authors
- Validating skills that rely on host-specific extensions, which the skill itself says are outside the portable contract
Try asking
- “Review my proposed skill folder at ./skills/invoice-checker before I publish it”
- “Check whether my SKILL.md frontmatter is valid and tell me what to fix”
- “Should this repeated workflow be a skill, a tool, or a scheduled step?”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 5 original files, the MIT licence and an attribution note.
skill-contract-reviewer.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip skill-contract-reviewer.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip skill-contract-reviewer.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Give the agent the absolute path of the installed skill folder and of the workspace so it resolves paths correctly.
- Ask for the execution evidence (script path, target path, cwd, argv, exit code) to confirm what was actually run.
- Treat a passing validation as a format check only; review any bundled scripts yourself before allowing execution.
What's inside
- assets/task-shapes.json
- references/contract.md
- references/decision-model.md
- scripts/check_skill.py
- SKILL.md
Redistributed unmodified from rohitg00/ai-engineering-from-scratch under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.